Re: [HTCondor-users] Can a DAGMan node be an already running job?

[Todd L Miller <tlmiller@xxxxxxxxxxx>]

> Let's say I have the following two DAGs
>
> A -> B
>
> A -> C
>
> Where A is a long running job. There's an A job running because someone 
> has invoked the A -> B workflow. Now someone comes along and wants to 
> invoke the A -> C workflow. I don't want to re-run A, but I can't mark 
> it as DONE in the DAGfile because then C will start running immediately.

	As far as I know, there's no particular time limit on how long a 
DAG node's PRE script can take, so if you know which job(s) or cluster(s) 
you're waiting for, you can wait for them in the PRE script.  That has two 
advantage: first, it's not stepping outside of DAGMan, meaning that as the 
workflow(s) involved get more complicated, you can continue to take 
advantage of (rather than reimplement) its features; second, it doesn't 
use an execute node while it's waiting.

	Of course, you still have to skip the job, but that's what the 
"skip_if_dataflow" submit command is for.  (There are other ways to skip 
the job from the PRE script, but this one has the advantage of having 
HTCondor verify that the two A nodes were, in fact, identical.)

-- ToddM